PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT EXCHANGERS USING AN EDUCATIONAL APPLICATION

Fundamental J. Thermal Science and Engineering, Vol. 2, Issue 1, 2012, Pages 37-52 Published online at http://www.frdint.com/ PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT EXCHANGERS USING AN EDUCATIONAL APPLICATION JOAQUÍN ZUECO a, SEYEDALI VEDAD b, MOHAMMADALI GHORBANI b, NILOOFAR HEIDARI b, MEHRNOOSH ZEINALKHANI b and ALIREZA HEIDARI b,* a ETS Ingeniería Industrial Departamento de Ingeniería Térmica y Fluidos Campus Muralla del Mar Universidad Politécnica de Cartagena 30202 Cartagena (Murcia), Spain b Institute for Advanced Studies Tehran 14456-63543, Iran e-mail: Prof.Alireza.Heidari@InstituteforAdvancedStudies.us Abstract Keywords and phrases: shell and tube, heat exchanger, LMTD method, effectiveness-ntu method. The shell and tube heat exchangers are devices used very much in the industrial processes in many applications. These elements can have characteristics very special, in its use as well as in its design, therefore each system being a particular problem. The program developed uses simultaneously the LMTD and the effectiveness-ntu methods to such important heat transfer devices and may prove pedagogically useful to undergraduate mechanical engineering students attending heat transfer courses. The examples in this paper illustrate the power of the software developed in thermal sciences calculations with application to the heatexchanger in configuration on line, staggered arrangement and of tube double. It is hoped that the software will bridge the gap between the * Corresponding author Received January 11, 2012 2012 Fundamental Research and Development International

38 JOAQUÍN ZUECO et al. teaching of engineering fundamentals and the existing industry practice of shell and tube heat exchanger design. 1. Introduction The demand for educational software is growing exponentially with the surge of interest in educational reform, the internet, and distance learning. Educational applications must be very flexible because curricula and teaching styles vary tremendously among institutions, locations, and even among instructors at the same institution. In the classroom, educators and students have an investment in traditional monolithic applications that can bias them against adopting new technologies. A lot has been written about designing heat exchangers, and specifically, shell-and-tube heat exchangers. For example, the book by Kern [1] published in 1950 details basic design procedures for a variety of heat exchangers. Since the publication of that book, with the advent of computers, design procedures have become sophisticated even though the basic goals of design remain the same. This paper presents a didactical software developed for the design of shell and tube heat exchangers. Shelland-tube heat exchangers (STHEs) are used extensively throughout the process industry and as such a basic understanding of their design, construction and performance is important to the practicing engineer. They are commonly used in process applications and in the air conditioning and refrigeration industry. They are also widely used as oil coolers, power condensers, preheaters and steam generators in both fossil fuel and nuclear-based energy production applications. They are also widely used. Although they are not specially compact, their robustness and shape make them well suited for high pressure operations. The design of a heat exchanger involves initial conditions in which the following variables are known: (i) process-fluid rate of flow (ii) change in temperature of process fluid (iii) inlet temperature of utility fluid (for cooling or heating). With this information, the engineer must prepare a design for the optimum exchanger that will meet the required process conditions. Ordinarily, the following results must be determined: (i) heat-transfer area (ii) exit temperature and flow rate of utility fluid (iii) number, length, diameter, and arrangement of tubes (iv) tube-side and shell-side pressure drops and (v) efficiency. To make the most of exchanger design software, one needs to understand shelland-tube heat exchangers (STHEs) classification, exchanger components, tube layout, baffling, pressure drop, and mean temperature difference. However, a

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 39 previous good understanding of the underlying principles of exchanger design is needed to use the software effectively. The software called INTERCAM, was developed to be used in undergraduate classes of Mechanical and Chemical Engineering. The use of the software facilitates the development of individual or group projects, in which the situation can develop its creativity through the scanning of a real problem. The software is suitable for teaching thermal design of shell and tube heat exchangers to senior undergraduate students in mechanical and chemical engineering and also to train new graduate engineers in thermal design. Students may be asked to use the software to undertake a few mini projects in shell and tube heat exchanger design. The software allows changing operation and design conditions such as temperature and flow inlet and outlet in shells and tubes, number and length of tubes, inside and outside diameter of tubes, number of passes, shell outside diameter, heat transfer coefficients in shells and tubes, heat transfer coefficient global, velocities of the flows and fouling factor. The three examples studied in this work illustrate how the software can be used in such a manner. The present work shows the software structure developed for teaching of the shell and tubes heat exchangers. INTERCAM is very easy to use and user friendly. It is a software tool that is made to demonstrate the thermal design calculations of shell and tube heat exchangers. The program shows the static behavior of these devices under different design parameters and operations conditions. The software allows changing operation and design conditions such as temperature and flow inlet in shells and tubes, number and length of tubes, inside and outside fouling factor. 2. Development of the Software: INTERCAM The tutorial model has been developed to apply in the teaching of shell and tube heat-exchanger, programmed in C + +. This software has main objective to solve problems to the heat-exchanger in configuration on line, staggered arrangement and of tube double. The software is divided in two sections. The first section is employed to introduce the data of the problem or charger previously a problem. The second section corresponds to the application to solve the problems. The use of this software permits to the actors of the learning teaching process (students and teachers): to realize analysis and synthesis of the process that including in operations with heat-exchanger, to apply the concepts and theory knowledge of

40 JOAQUÍN ZUECO et al. various subjects in the practical of laboratory, to learn of manner autonomy and to incorporate new information and to decide the form more adequate use of it. The students can do practical in the laboratory with a real system and/or simulations in the software with data found in the literature. This methodology can be applied in second/third of a bachelor course. The software for education is a tool useful to get: class more active, problems with solutions different, programming of tasks of analysis and others formative actions. The use of this type of tools has been developed and proposed by various authors [2-6] developed a software structure for teaching and learning of the dynamics and control of shell and tubes heat exchangers. This software allows changing operation and design conditions such as temperature and flow inlet in shells and tubes, number and length of tubes, inside and outside diameter of tubes, number of passes, shell outside diameter and fouling factor. [7] introduced the effect of temperature on fouling into the design of heat exchangers, being quite complex because the final, design-fouling resistance depends on the history of operation of the exchanger. In the recent past, some experts studied on the design, performance analysis and simulation studies on heat exchangers [8-11]. It is possible to find two possibilities: development of educational software and use of commercial software. An overall skeleton of the menu, which is designed to be very user-friendly, is shown in Figure 1. There are three command boxes, namely Initial data, Configuration and Results. Initial data presents the values of the problem load. Configuration is possible to change the data, besides of design a new problem or to load other problem previously designed (Figure 2). Finally, the values calculated for all the parameters are shown in results. Figure 3 shows (see [12]) an example of a type of heat exchanger, a fixed-tube sheet heat exchanger. This device has straight tubes that are secured at both ends to tube sheets welded to the shell. The principal advantage of the fixed tube sheet construction is its low cost because of its simple construction. In fact, the fixed tube sheet is the least expensive construction type, as long as no expansion joint is required. Other advantages are that the tubes can be cleaned mechanically after removal of the channel cover or bonnet, and that leakage of the shell side fluid is minimized since there are no flanged joints. A disadvantage of this design is that since the bundle is fixed to the shell and cannot be removed, the outsides of the tubes cannot be cleaned mechanically.

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 41 Figure 1. Main screen of the INTERCAM program. Figure 2. Screen of configuration of parameters. Figure 3. Example of shell-and-tube heat exchanger [8].

42 JOAQUÍN ZUECO et al. 3. Educational Objectives A heat exchanger is a device in which energy is transferred from one fluid to another across a solid surface. Exchanger analysis and design therefore involve both convection and conduction. Two important problems in heat exchanger analysis are (i) rating existing heat exchangers and (ii) sizing heat exchangers for a particular application. The shell and tube heat exchanger is equipment largely used at chemical industries. The developed software is intended to be used by undergraduate mechanical and chemical engineering students at the second year level. It can therefore be easily appreciated that the developed software has versatile teaching and educational uses. The software can be a useful tool: (i) To introduce the student to the basic concepts of heat-exchanges. (ii) To enable the student to interactively create and utilize different types of dispositions of this type of heat-exchanger. (iii) To make the student appreciate the effects of various pertinent-actors such as global and convective heat transfer coefficients, efficiency, shell and tube passes, shell diameter, etc. (iv) To qualify the student to interpret the results obtained. (v) To enable the student to learn the fundamentals of this type of device. 4. Mathematical Model The LMTD method can be readily used when the inlet and outlet temperatures of both the hot and cold fluids are known. When the outlet temperatures are not known, the LMTD can only be used in an iterative scheme. In this case the effectiveness- NTU method can be used to simplify the analysis. The basic heat exchanger equations applicable to shell and tube exchanger are shown here: Total heat load transfer q = U A Tm F, (1) o where U is the global heat transfer referred to the area A o, coefficient that depends that the heat transfer coefficients (inside and outside), geometry of the exchanger and thermal properties of both fluids. A is the heat transfer area, cc Tmcc is the

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 43 logarithmic mean temperature difference for the pure countercurrent flow configuration and F is the configuration correction factor for multiple tube-side and/or shell-side passes. U A o 1 =, (2) Ao Ao ln( do di ) 1 + + A h 2πkL h i i o = 2πr L, (3) o ( Thi Tco ) ( Tho Tci ) Tmcc =, (4) ( Thi Tco ) ln ( T T ) ho where T hi and T ci are the hot and cold inlet temperatures, respectively, T ho and T co are the corresponding outlet temperatures. The value of F depends upon the exact arrangement of the streams within the exchangers, the number of exchangers in series, and two dimensionless parameters defined in terms of the terminal temperatures of the two streams: ci ( Thi Tho ) R = ( T T ), (5) co ci ( Tco Tci ) P = ( T T ). (6) ho The mathematical relationships between R, P and F can be found in numerous books [1, 13], we have considered in this software the configurations more used, 1-2 and 2-4. Values of F below 0.75 at the lowest should not be used, because it means that substantial additional area must be supplied in the heat exchanger to overcome the inefficient thermal profile. F is unity for the cases of purely parallel and countercurrent flow. The Arithmetic Mean Temperature Difference is the following in this case. The expression of ci ( Thi Tci ) ( Tho Tco ) Tm p =. (7) ( Thi Tci ) ln ( T T ) ho co Tm for the case purely concurrent (parallel) is rarely used. The LMTD method and NTU method are used in the software developed. The NTU method is an alternative to calculate the total heat transfer rate. This method brings

44 JOAQUÍN ZUECO et al. out the individual effects of not only the total thermal conductance UA but also that the capacity rates denominated number of heat transfer units: where C h and C c, so appears a new dimensionless number, C min is the smaller of the two capacity rates. NTU = UA C min, (8) Other dimensionless number is the heat exchanger effectiveness ε, which is defined as the ratio between the actual heat transfer rate q and the maximum heat transfer that could take place in the heat exchanger ( q = C T ). Heat-transfer coefficient max min max ε = q q max. (9) The tube side heat-transfer coefficient is a function of the physical properties (viscosity, thermal conductivity and specific heat), the tube diameter and very importantly, mass velocity. It is necessarily calculated the Reynolds number and the Prandtl number. The variation in liquid viscosity is quite considerable; so, this physical property has the most dramatic effect on heat-transfer coefficient. The fundamental equation for turbulent heat-transfer inside tubes is:.8 0.33 Nu = 0.027Re 0 Pr. (10) The models derived above are simulated using the software developed in this work and called INTERCAM. To built this program, it was necessary to use the programming Language C ++. 4. Results and Discussions Usually, the flow rates and the physical properties of the two streams involved are specified, and the temperatures at which the fluids are available are known. If the outgoing temperature of one of the streams is not specified, usually a constraint (e.g., the temperature of the cooling water cannot exceed 99 C ) is given. Then, by an energy balance, the outgoing temperature of the second stream can be calculated along with the heat duty. The heat duty q is usually fixed by the required service. The tube diameter, tube length, shell types, etc. are all standardized and are available only in certain sizes and geometry. The selected heat exchanger has to meet or

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 45 exceed this requirement. Problem I Configuration: 1-2 (one tube pass and two shell passes) and 2-4 (two tube passes and four shell passes), double tube. Hot fluid (shell side): water air NH3 oil, 1.19 kg s, T hi = 230 C. Cold fluid (tube side): water, 0.72 kg s, T ci = 30 C. 2 Global heat transfer coefficient: U = 1192 W m K. Geometric parameters: d i Length of the tube: 2.5 m. = 40 mm. Table 1. Problem I. Configuration 1-2, double tube for various hot fluids Hot fluid q ( kw) c ( kj KgK ) NTU ε Tm e, h Water 124.12 4.180 0.249 0.206 166.77 Air 89.96 1.031 1.009 0.6065 123.25 NH 3 114.98 2.467 0.422 0.324 155.15 Oil 125.64 4.689 0.222 0.186 168.70 The effect of different input variables on output variable are discussed in detail in the following cases. Problem I is studied in the Tables 1-3. Tables 1 and 2 show the effect of the change of the hot fluid (water, air, NH 3 and oil) for the configurations 1-2 and 2-4, respectively. When the hot fluid chosen is the air, the biggest values of NTU and ε are obtained, however q decreases. We can see that the best configuration is the 2-4, because for this configuration are reached the performance maximum. Table 2. Problem I. Configuration 2-4, double tube. Effects of the hot fluid Hot fluid q ( kw) c ( kj KgK ) NTU ε Tm e, h Water 326.29 4.180 0.498 0.542 111.63 Air 142.65 1.031 2.018 0.961 52.98 NH 3 262.86 2.467 0.483 0.843 91.48 Oil 337.81 4.689 0.444 0.500 115.27

46 JOAQUÍN ZUECO et al. Table 3 shows the effects of inlet hot fluid temperature. It can be seen that an increase in the T hi leads to a decrease in heat exchanger effectiveness and NTU. Besides, the actual heat transfer rate increases as the inside hot fluid temperature increases. Table 3. Problem I. Configuration 1-2, double tube. Effects of inlet hot fluid temperature T ( C) ε NTU q ( kw) hi Problem II Tm 220 0.607 1.011 85.39 117.01 200 0.608 1.015 76.29 104.54 180 0.609 1.018 67.23 92.14 160 0.610 1.021 58.20 79.76 Configuration: in-line tube banks (Figure 4a). Hot fluid (shell side): water, 3.55 kg s, T = 120 C, u = 0.5 m s. Cold fluid (tube side): oil, 5 kg s, u c = 0.4 m s, T ci = 30 C, T co = 85 C. 2 Global heat transfer coefficient: U = 1550 W m K. hi h Geometric parameters: d i = o 20 mm, d = 25 mm. With this parameters are obtained the following constant values for all the cases studied: N = 46, ε = 0.611, NTU = 1.630, c pf = 2 kj kgk, c pc = 4.18 kj kgk, q = 550 kw. Problem II is studied in the Tables 4 and 5. In Table 4 is used the 1-2 configuration, while in Table 5 is used the 2-4 configuration. In this case the effect of the global heat transfer coefficient is analyzed. An increase of this parameter leads to a decrease in the heat transfer and of the length of the tubes, which was expected. However, ε and NTU does not change. As in the previous case, the best configuration is the 2-4, because for this configuration are obtained less values of the heat transfer area and of the length of the tubes with the same performance. Table 4. Problem II. Configuration 1-2, in-line tube banks. Effects of the global heat transfer coefficient ( W m 2 2 U K ) A ( m ) L ( m) ε NTU 1250 13.039 1.8045 0.611 1.630 1350 12.073 1.6708 0.611 1.630 1450 11.240 1.5556 0.611 1.630 1550 10.515 1.4552 0.611 1.630

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 47 Table 5. Problem II. Configuration 2-4, in-line tube banks. Effects of the global heat transfer coefficient ( W m 2 2 U K ) A ( m ) L ( m) ε NTU 1250 10.651 0.7370 0.767 1.331 1350 9.862 0.6824 0.767 1.331 1450 9.182 0.6354 0.767 1.331 1550 8.589 0.5944 0.767 1.331 Problem III Configurations: 1-2, in-line and staggered tube banks (Figure 4). Hot fluid (shell side): water, 3.3 kg s, T hi = 100 C, Tho = 70 C. Cold fluid (tube side): water, 5.1kg s, 0.3 m s, T ci = 35 C. Thermal conductivity of the tube: k = 100 W mk. Geometric parameters: d i = o Baffle spacing: S b = 0.2 m. X 1 = 0.075 m, X 2 = 0.05 m. 20 mm, d = 25 mm. (a) Figure 4. Configurations: (a) Configuration in-line tube banks, (b) Configuration staggered tube banks. (b)

48 JOAQUÍN ZUECO et al. Table 6. Problem III. Configuration in-line tube banks. Effects of the distance between deflectors S b( m) h c ( W mk ) L ( m) ( 2 A m ) Φ s ( m) U ( W m 2 K ) 0.10 1886.9 1.319 11.396 0.3725 968.4 0.15 1461.5 1.516 13.098 0.4169 842.5 0.20 1219.3 1.689 14.598 0.4570 755.9 0.25 1059.4 1.848 15.964 0.4938 691.2 0.30 949.4 1.994 17.232 0.5281 640.4 The effect of the baffle spacing is analyzed in Table 6. It can be appreciated that when S b increases, A, L and the shell diameter ( Φ s ) increases, while h c (heat transfer coefficient for cold fluid or tube side) and U (global heat transfer coefficient) decreases. In Table 7 is studied the effect of the number and distribution ( column row ) of the tubes. It can be seen that an increase of N deals a decrease of the velocity, of the length tube and of the global heat transfer coefficient. The shell diameter ( Φ s ) increases when N increases, however, Φ s is 0.4773 and 0.4718 for N = 60 ( 12 columns 5rows) and N = 66 ( 11columns 6 rows), respectively. Table 7. Problem III. Configuration staggered tube banks. Effects of the number and distribution of the tubes N ( m s ) u f ( m) L A ( m ) 2 Φ ( m) U ( W m 2 K ) 50 ( 10 5) 0.328 1.6337 12.831 0.4357 860.0 60 ( 10 6) 0.273 1.4475 13.643 0.4498 808.9 60 ( 12 5) 0.273 1.5415 14.529 0.4773 759.6 66 ( 11 6) 0.248 1.4051 14.568 0.4718 757.5 Tables 8 and 9 show the effect of X 1 (vertical distance between tubes) for the configurations in-line and staggered tube banks, respectively. It can be seen that an increase of X 1 leads to a decrease in global heat transfer coefficient (U) and heat transfer coefficient h c. Besides, L and A increase when X 1 increases. We can see that for the staggered tube banks configuration are obtained bigger values of U and h c than the in-line tube banks configuration. Besides, the values of L and A are smaller, therefore the staggered tube banks configuration is the most adequate. s

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 49 Table 8. Problem III. Configuration in-line tube banks. Effects of the vertical distance between tubes X 1( m) h c( W mk) L ( m) ( 2 A m ) Φ s ( m) U ( W m 2 K ) 0.070 1302.9 1.6224 14.017 0.4414 787.3 0.075 1219.3 1.6897 14.598 0.4720 755.9 0.080 1148.2 1.7546 15.158 0.4570 728.0 Table 9. Problem III. Configuration staggered tube banks. Effects of the vertical distance between tubes X 1( m) h c ( W mk) L ( m) ( 2 A m ) Φ s ( m) U ( W m 2 K ) 0.070 1428.7 1.5362 13.271 0.4414 831.5 0.075 1359.8 1.5815 13.663 0.4720 807.7 0.080 1300.9 1.6240 15.158 0.4570 786.5 Table 10. Problem III. Configuration in-line tube banks. Effects of the inlet cold fluid temperature T ( C) F L ( m) NTU ε ci 25 0.96 1.3618 0.624 0.400 731.2 30 0.95 1.5029 0.700 0.429 744.0 35 0.94 1.6897 0.800 0.462 755.9 40 0.91 1.9497 0.910 0.500 766.8 45 0.88 2.3366 1.137 0.545 777.0 50 0.81 2.9461 1.477 0.600 786.4 Finally, Table 10 shows the effect of U ( W m 2 K ) T ci (inlet cold fluid temperature) in the configurations in-line. It can be observed that an increase in T ci leads to an increase in the heat transfer efficiency ( ε ), global heat transfer coefficient (U), transmission unities number (NTU) and the length tube (L), while the LMTD correction factor (F) decreases when T ci increases. Conclusion This software has been developed for the thermal design of shell and tube heat exchangers based on the LMTD and efficiency methods (NTU). The software is highly user-friendly and has many capabilities to create new problems in the field of the mechanical engineering in the design of shell-tube heat-exchanger. It is hoped that the software will bridge the gap between the teaching of fundamentals and industry practice of this type of device. The software incorporate various design

50 JOAQUÍN ZUECO et al. options for the heat exchangers including the variations in the tube diameter, tube pitch, shell diameter, number of tube passes, baffle spacing, baffle cut, etc. Three configurations can be used, on line, staggered arrangement and of tube double. The software was tested using some thermal problems and found to provide results that are very accurate. Trial tests show that INTERCAM solves the problems directly and effectively using an ordinary PC. Nomenclature 2 A [ m ] Heat-transfer area c [ kj kgk] Specific heat e d i [ m ] Tube inside diameter d o [ m ] Tube outside diameter F [ ] LMTD correction factor h [ W m 2 K ] Heat-transfer coefficient k [ W mk] Thermal conductivity L [ m ] Length tube q [ KW ] Heat transfer rate ṁ [ kg s] Flow rate N [ ] Number of tubes NTU [ ] Transmission unities number Nu [ ] Nusselt number Pr [ ] Prandtl number P [ ] Dimensionless parameter, LMTD R [ ] Dimensionless parameter, LMTD R [ m 2 K kj ] Fouling resistance f

PERFORMANCE ANALYSIS OF SHELL AND TUBE HEAT 51 Re [ ] Reynolds number T [ C] Temperature u [ m s] Velocity U [ W m 2 K ] Global heat transfer coefficient S b [ m ] Baffle spacing ε [ ] Efficiency µ [ kg m s] Viscosity Φ s [ m ] Shell diameter Special characters Subscripts c h i o s t Cold Hot Inlet Outlet Shell Tube References [1] D. Q. Kern, Process Heat Transfer, McGraw-Hill, New York, 1950. [2] H. I. Abu-Mulaweh, Experimental comparison of heat transfer enhancement methods in heat exchangers, Int. J. Mech. Engrg. Edu. 31 (2003), 160-167. [3] C. K. Leong, K. C. Toh and Y. C. Leong, Shell and tube heat exchanger design software for educacional applications, Int. J. Engrg. Edu. 14 (1998), 217-224. [4] L. M. F. Lona, F. A. N. Fernandes, M. C. Roque and S. Rodrigues, Developing an educational software for heat exchangers and heat exchanger networks projects, Computers & Chemical Engineering 24(2-7) (2000), 1247-1251.

52 JOAQUÍN ZUECO et al. [5] F. L. Tan and S. C. Fok, An educational computer-aided tool for heat exchanger design, Computer Applications in Engineering Education 14 (2006), 77-89. [6] F. Machuca and O. Urresta, Educational software for the teaching of the dynamics and control of shell and tube heat exchangers, Rev. Fac. Ing. Univ. Antioquia 44 (2008), 52-60. [7] D. Butterworth, Design of shell-and-tube heat exchangers when the fouling depends on local temperature and velocity, Appl. Thermal Engineering 22 (2002), 789-801. [8] M. T. Aghareed, M. A. El-Rifai, Y. A. El-Tawil and R. M. Abdel-Monen, A new dynamic model for shell and tube heat exchangers, Energy Conservation Management 32 (1991), 439-446. [9] S. Anantharaman, Design and construction of shell and tube heat exchangers, Chem. Ind. Dev. Incorporat. Chemical Processing Engineering 11 (1997), 15-21. [10] S. A. Mandavgane, M. A. Siddique, A. Dubey and S. I. Pandharipande, Modeling of heat exchangers using artificial neural network, Chemical Engineering World, 2004, pp. 75-80. [11] A. K. Yusuf and O. Guraras, A computer program for designing of shell and tube heat exchangers, Appl. Thermal Engineering 24 (2004), 1797-1805. [12] R. Mukherjee, Effectively design shell-and-tube heat exchangers, Chemical Engineering Progress, 1998. [13] A. Bejan, Heat Transfer, John Wiley & Sons, Inc., New York, 1993.